Multi-Dimensional Resource Optimal Allocation Scheme for Tiansuan Constellation

With the advent of the 6G era, the construction of space networks is gradually entering a peak period. Tiansuan constellation came into being in this context, which aims to provide an open research platform for researchers. Tiansuan constellation can be used as the coverage supplement of terrestrial networks and the traffic supplement in hotspot areas, providing flexible, efficient, and seamless coverage of humanized services for global users. Satellite network resource management is more difficult than terrestrial network resource management due to the heterogeneity and difference in characteristics of satellite and terrestrial networks. This paper proposes a multi-dimensional network resource allocation algorithm for the Tiansuan constellation. Considering the limited storage resources and bandwidth resources of the satellite Internet, and taking into account the computing resources of the satellite Internet, the joint optimization of multi-dimensional resources is realized. The policy network-based reinforcement learning model is adopted to independently optimize the decision-making process of satellite Internet resource allocation. Compared with the two baseline algorithms, the proposed algorithm improves the network resource allocation profit and user service rate by 29.9% and 10.7%, respectively. In addition, the effectiveness and flexibility of the proposed scheme are verified by adjusting the storage resource requirements of users.